Assessment of the accuracy of multiple sonographic fetal weight estimation formulas: a 10-year experience from a single center.

Objectives-The primary aim of this study was to compare the accuracy of sonographic fetal weight estimation models. The secondary aim was to define the most accurate time (4-7 or 3 days before delivery) for evaluating fetal weight. Methods-In this retrospective cohort study, a total of 12,798 sonographic fetal weight estimations were analyzed, of which 9459 were performed within 3 days of delivery and 3339 within 4 to 7 days. The cohort included all singleton pregnancies recorded at a single medical center from January 2000 to December 2010, with 24 weeks' gestation minimum. Predicted birth weights were calculated according to 23 sonographic fetal weight estimation models; in total, 294,354 sonographic weight estimations were evaluated and compared to the actual birth weights. Results-The accuracy of the models in predicting birth weight differed considerably. The most accurate models used 3 or more fetal measurements followed by models using abdominal circumference only. The models developed by Sabbagha et al (Am J Obstet Gynecol 1989; 160:854-862) proved most accurate, with a mean percent error of-0.2% and greater than 92% of estimates within 15% of birth weight (P<.05). Nineteen sonographic fetal weight estimation models (82.6%) better predicted fetal weight at 4 to 7 days before delivery (P<.001). Twenty-two (95%) of the models were less accurate at the extreme ends of fetal weight. Conclusions-Different formulas for fetal weight estimation vary greatly; we recommend that each center should evaluate the most accurate formula according to its attending population. Estimation of fetal weight performed 4 to 7 days before delivery using most models was more accurate than estimations performed 3 days before delivery.